Stepwise approach for combining many sources of evidence for site-recognition in genomic sequences

BMC Bioinformatics

Table 4 The table shows the specificity (Sp), sensitivity (Sn), true positives (TP), true negatives (TN), false negatives (FN), false positives (FP) and area under the ROC and PRC curves (auROC/PRC) as well as the time needed for obtaining the best combination for all methods and the five studied chromosomes for TIS recognition

Chromosome	Objective	Method	Combination	auROC	auPRC	G	Sp	Sn	TP	FN	TN	FP	#models	Time (s)
Chr. 1	-	Std	-	0.9473	0.0929	0.8430	0.9528	0.7458	1608	548	7,693,177	381,413	-	-
	auROC	Cons	Sum	0.9820	0.1529	0.8893	0.9874	0.8010	1727	429	7,972,989	101,601	8	1555
	auPRC	Cons	Sum	0.9669	0.1859	0.8326	0.9859	0.7032	1516	640	7,960,941	113,649	29	3436
	G	Cons	Majority	0.9445	0.0074	0.9353	0.9599	0.9114	1965	191	7,750,802	323,788	2	572
Chr. 3	-	Std	-	0.9334	0.0831	0.8335	0.9584	0.7248	843	320	6,988,324	303,627	-	-
	auROC	Cons	Sum	0.9761	0.1449	0.8458	0.9929	0.7206	838	325	7,240,060	51,891	8	1807
	auPRC	Cons	Sum	0.9596	0.1762	0.7891	0.9920	0.6277	730	433	7,233,286	58,665	33	3708
	G	Cons	Majority	0.9315	0.0049	0.9238	0.9664	0.8831	1027	136	7,046,822	245,129	2	696
Chr. 13	-	Std	-	0.9506	0.0844	0.8692	0.9479	0.7971	271	69	3,473,147	191,017	-	-
	auROC	Cons	Sum	0.9778	0.1347	0.8429	0.9940	0.7147	243	97	3,642,333	21,831	7	1348
	auPRC	Cons	Sum	0.9692	0.1608	0.7924	0.9929	0.6324	215	125	3,638,155	26,009	28	3280
	G	Cons	Majority	0.9483	0.0038	0.9385	0.9630	0.9147	311	29	3,528,645	135,519	2	476
Chr. 19	-	Std	-	0.9456	0.1277	0.8727	0.9068	0.8398	1185	226	1,540,551	158,340	-	-
	auROC	Cons	Sum	0.9716	0.1726	0.8876	0.9691	0.8129	1147	264	1,646,337	52,554	7	1510
	auPRC	Cons	Sum	0.9575	0.2048	0.8534	0.9657	0.7541	1064	347	1,640,628	58,263	27	3269
	G	Cons	Majority	0.9436	0.0181	0.9346	0.9555	0.9142	1290	121	1,623,271	75,620	2	473
Chr. 21	-	Std	-	0.9353	0.0726	0.8409	0.9415	0.7511	175	58	1,227,429	76,205	-	-
	auROC	Cons	Sum	0.9809	0.1299	0.8612	0.9875	0.7511	175	58	1,287,377	16,257	8	1526
	auPRC	Cons	Sum	0.9670	0.1635	0.8201	0.9857	0.6824	159	74	1,284,991	18,643	31	3352
	G	Cons	Majority	0.9362	0.0043	0.9271	0.9582	0.8970	209	24	1,249,122	54,512	2	288

The results of our method for the three different evaluation measures are shown in boldface

ISSN: 1471-2105